the first-perspective alignment effect: the role of environmental complexity and familiarity with...

This article was downloaded by: [University of California, San Francisco]On: 30 September 2014, At: 01:14Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office:Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

The Quarterly Journal of ExperimentalPsychologyPublication details, including instructions for authors and subscriptioninformation:http://www.tandfonline.com/loi/pqje20

The first-perspective alignment effect:The role of environmental complexity andfamiliarity with surroundingsMichael Tlauka a , Pelham Carter b , Tim Mahlberg a & Paul N. Wilson ba School of Psychology , Flinders University , Adelaide , Australiab Department of Psychology , University of Hull , Hull , UKAccepted author version posted online: 24 May 2011.Published online: 15Aug 2011.

To cite this article: Michael Tlauka , Pelham Carter , Tim Mahlberg & Paul N. Wilson (2011) The first-perspective alignment effect: The role of environmental complexity and familiarity with surroundings, TheQuarterly Journal of Experimental Psychology, 64:11, 2236-2250, DOI: 10.1080/17470218.2011.586710

To link to this article: http://dx.doi.org/10.1080/17470218.2011.586710

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)contained in the publications on our platform. However, Taylor & Francis, our agents, and ourlicensors make no representations or warranties whatsoever as to the accuracy, completeness, orsuitability for any purpose of the Content. Any opinions and views expressed in this publicationare the opinions and views of the authors, and are not the views of or endorsed by Taylor &Francis. The accuracy of the Content should not be relied upon and should be independentlyverified with primary sources of information. Taylor and Francis shall not be liable for anylosses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilitieswhatsoever or howsoever caused arising directly or indirectly in connection with, in relation to orarising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantialor systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, ordistribution in any form to anyone is expressly forbidden. Terms & Conditions of access and usecan be found at http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/loi/pqje20

http://www.tandfonline.com/action/showCitFormats?doi=10.1080/17470218.2011.586710

http://dx.doi.org/10.1080/17470218.2011.586710

http://www.tandfonline.com/page/terms-and-conditions

The first-perspective alignment effect: The role ofenvironmental complexity and familiarity with

surroundings

Michael Tlauka1, Pelham Carter2, Tim Mahlberg1, and Paul N. Wilson2

1School of Psychology, Flinders University, Adelaide, Australia2Department of Psychology, University of Hull, Hull, UK

People often remember relatively novel environments from the first perspective encountered or the firstdirection of travel. This initial perspective can determine a preferred orientation that facilitates theefficiency of spatial judgements at multiple recalled locations. The present study examined this “first-perspective alignment effect” (FPA effect). In three experiments, university students explored three-path routes through computer-simulated spaces presented on a desktop computer screen. Spatialmemory was then tested employing a “judgement of relative direction” task. Contrary to the predictionsof a previous account, Experiment 1 found a reliable FPA effect in barren and complex environments.Experiment 2 strongly implicated the importance of complete novelty of the space surrounding theroute in producing the effect. Experiment 3 found that, while familiarity with the surrounding spacegreatly attenuated the FPA effect with immediate testing, the effect reemerged following a 7-daydelay to testing. The implications for the encoding and retrieval of spatial reference frames are discussed.

Keywords: Frame of reference; Egocentric learning; Alignment; Spatial memory.

Whether finding our way through a novel city ormoving around our familiar home, spatialmemory is a basic but vital cognitive competency.A common assumption (e.g., Burgess, 2006) isthat spatial memories are organized around twoprimary frames of reference: egocentric and allo-centric. Sholl and Nolin (1997; see also Easton &Sholl, 1995) suggest that as people move throughan environment, egocentric self-to-object relationsare updated in a body-reference system that relieson the three principal axes of head/feet, front/back, and left/right (Franklin & Tversky, 1990).

The spatial interrelationships between objects inthe environment are stored as an allocentric,object-to-object representation, providing a globalreference frame of stable environmental features(Mou, McNamara, Valiquette, & Rump, 2004;Waller & Hodgson, 2006).

There is evidence that the initial experience in anovel environment determines preferential spatialrecall (e.g., Shelton & McNamara, 2001; Wilson,Tlauka, & Wildbur, 1999). However, recently,Mou, Zhao, and McNamara (2007) found that insome cases (in the presence of a strong intrinsic

Correspondence should be addressed to Michael Tlauka, School of Psychology, Flinders University, GPO Box 2100, Adelaide

SA5001, Australia. E-mail: [email protected]

2236 # 2011 The Experimental Psychology Society

http://www.psypress.com/qjep http://dx.doi.org/10.1080/17470218.2011.586710

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY

2011, 64 (11), 2236–2250

Dow

nloa

ded

by [

Uni

vers

ity o

f C

alif

orni

a, S

an F

ranc

isco

] at

01:

14 3

0 Se

ptem

ber

2014

structure), initial views had no influence on learn-ing, and the authors question whether the firstencountered perspective has a critical role in refer-ence frame selection. The current investigationfurther examines whether the first perspective onan environment establishes a preferred frame ofreference. More specifically, the focus of thecurrent investigation is on the conditions (environ-mental complexity and familiarity with surround-ings) that promote a frame of reference that istied to the first perspective or first direction oftravel within a novel, large-scale environment.

Wilson et al. (1999) asked participants to read orlisten to a description of a right-angled U-shapedroute (see Figure 1) prior to making judgementsof relative direction (JRD) of the form “Imagineyou are at A, facing B. Point toward C”.Estimates from imagined perspectives that were inalignment with the first part of the route that par-ticipants encountered resulted in lower error scores(and faster latencies) than scores from those misa-ligned with the initial orientation. ConsiderFigure 1: Judgements from memory based on theroute section A to B (the initially experienced per-spective) as well as those from D to C (a segment

aligned with the original perspective) were moreefficient than misaligned judgements (e.g., B toC) and contra-aligned judgements (e.g., C to D).This phenomenon also occurs following real-world exploration and is referred to as the first-per-spective alignment (FPA) effect (Wilson, Wilson,Griffiths, & Fox, 2007). Wilson and Wildbur(2004) reported similar results after participantslearned a return journey along a U-shaped path ina computer-generated virtual environment ofsimilar shape to that employed by Wilson et al.(1999; see Richardson, Montello, & Hegarty,1999; Rossano, West, Robertson, Wayne, &Chase, 1999).

The FPA effect refers to superior judgements forviewpoints that are aligned with the orientation ofthe initial perspective. There are at least three ques-tions of interest with respect to the FPA effect. Thefirst relates to its generalizability. Apart from verbaldescriptions (Wilson et al., 1999) and virtual learn-ing (Richardson et al., 1999; Rossano et al., 1999;Wilson & Wildbur, 2004), alignment with thefirst perspective has been found following blind-folded exploration of a real-world small-scaleroute (Palij, Levine, & Kahan, 1984) and after pre-sentations of multiple perspectives of a real-worldarray of objects (Shelton & McNamara, 2001,Experiment 7) and in real-world routes within cor-ridors and surrounding buildings (Wilson et al.,2007). In contrast, Mou et al. (2007) found no evi-dence for a benefit in headings aligned with the firstperspective.

The second question concerns the experimentalconditions that influence this form of learning.Wilson et al. (2007) reported that the FPA effectdiffered in strength depending on different aspectsof the environment—that is, the allocentric arrange-ments of features and landmarks (see also Kelly &McNamara, 2008; Shelton & McNamara, 2001).They speculated that the effect was influenced bythe number and arrangement of features externalto the route. When participants explored a real-world corridor (which occluded aspects of the sur-rounding environment), an FPA effect waspresent, but when participants explored a similarlystructured route around the outside of a building(with surrounding distal landmarks), the effect was

Figure 1. An illustration of the routes employed in Experiments 1–3.

Some aspects of the environment (the surrounding area, objects, and

targets) are not shown. In the example, participants started

exploration at Position A. The first-perspective alignment (FPA)

effect refers to superior judgements from the route section A to B

(the original perspective) as well as those from D to C (a segment

aligned with the original perspective).

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2011, 64 (11) 2237

FIRST-PERSPECTIVE ALIGNMENT

Dow

nloa

ded

by [

Uni

vers

ity o

f C

alif

orni

a, S

an F

ranc

isco

] at

01:

14 3

0 Se

ptem

ber

2014

attenuated.Wilson et al. argue that the primary per-spective on a novel environment establishes an“anchor point” (Couclelis, Golledge, Gale, &Tobler, 1987), which is used to organize spatialknowledge, assisting in navigation and orientationtasks. It is the salience of this anchor point relativeto other features in the environment that determinesmemory organization (rather than the arrangementof intrinsic or extrinsic cues in the environment assome models suggest, e.g., Kelly & McNamara,2008). Experiments 1 and 2 address these issues.

The third question relates to the stage of proces-sing responsible for FPA effects. The evidence fordiffering spatial frames of reference comes largelyfrom testing in which participants are asked torecall aspects of an explored or learned space. Oneor more perspectives are typically preferred (i.e.,more efficiently recalled). However, little isknown about whether or to what extent a preferredperspective is established during the recall task, orestablished at encoding during the course ofexploration. In other words, are spatial memorieslargely the result of factors occurring duringstorage (when information is consolidated inlong-term memory) or during retrieval (wheninformation is accessed from long-term memory)?Experiment 3 allowed us to begin to investigatethe question of whether a frame of reference isestablished primarily at encoding or recall.

In the three experiments reported here, partici-pants explored large-scale desktop virtual environ-ments that modelled urban buildings. Virtualenvironments (VEs) have been used extensively inspatial learning research, and it is generallyassumed that learning in real and computer-simulated space is mediated by similar cognitive pro-cesses (Waller, 2000; Wilson, Foreman, & Tlauka,1997; Witmer, Bailey, & Knerr, 1996; for differ-ences see Hegarty, Montello, Richardson,Ishikawa, & Lovelace, 2006).

EXPERIMENT 1

Experiment 1was designed to addressWilson et al.’s(2007) hypothesis based on their investigation,which found an FPA effect when people explored

an arrangement of corridors but a greatly attenuatedeffect when participants explored a similar route sur-rounding a building. Establishing the conditionsthat promote and attenuate an effect is crucial forinvestigating its genesis. Wilson et al. speculatedthat attenuation of the FPA effect only occurredwhen participants explored an outside environmentwith distal spatial features available to serve asalternative “anchor points” that could compete withthe primary anchor point established by experienceat the initial perspective. A secondary aspect to thishypothesis is whether alternative anchor pointsdepend on simply the number of alternative cues(cue richness or complexity), which could be variedwithin an otherwise restricted corridor-like environ-ment.Alternatively, the arrangement of cues could becrucial, requiring relatively distal cues or widespacing, as available in a typical “outside” environ-ment, but not within a corridor. It is also conceivablethat a particular combination of richness and spacingmight be important. Therefore, in Experiment 1 wevaried indoor/outdoor and richness/impoverishmentin a parametric design.

Four groups of participants explored a simple U-shaped route (Figures 1 and 2) that was based onprevious related research (Wilson et al., 1999;Wilson & Wildbur, 2004; Wilson et al., 2007).The environments differed on two dimensions:viewing range (indoor, outdoor) and level of detail(rich in features, impoverished). The participantsnavigated routes with identical dimensionsthrough an impoverished–indoor corridor VE, animpoverished–outdoor VE, a rich–indoor VE, ora rich–outdoor VE. Following exploration, partici-pants took part in a JRD task, which assessed theirspatial knowledge of the route. Our a priorihypothesis based on the outcome of Wilson et al.(2007) was that the FPA effect would be mostevident in the impoverished–indoor environment,whereas exploration of a rich–outdoor environmentwould lead to the greatest attenuation of the effect.

Method

ParticipantsThe participants were 96 undergraduate studentswho took part to fulfil a course requirement.

2238 THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2011, 64 (11)

TLAUKA ET AL.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

alif

orni

a, S

an F

ranc

isco

] at

01:

14 3

0 Se

ptem

ber

2014

Their mean age was 21 years (range: 18–58 years).The students were randomly allocated to fourgroups (n= 24).

ApparatusA desktop VE was created using the Hammereditor for the Half Life 2 Game Engine. Theenvironment was presented via an Intel PentiumDesktop computer with SVGA graphics, displayedon a 17-inch LCD monitor. Participants used thefour arrow keys/cursors to navigate. The up anddown keys controlled forward and backwardmotion, respectively. The left and right keyrotated the direction the participant was facing.

Participants practised virtual navigation prior tothe experimental tasks in an unrelated practice

environment with dimensions that correspond toapproximately 68 metres by 68 metres. Theenvironment was an enclosed space containingfour large objects to navigate through and around.It bore no resemblance to the experimentalenvironments.

The experimental environments were based onthe real-life outdoor environment used by Wilsonet al. (2007, Experiment 3). The rich–outdoorvirtual environment was the closest model to thereal environment used by Wilson et al. (seeFigure 2): Participants followed a path adjacent tothree of the exterior walls of an oblong building(36 metres by 12 metres). The VE containedseveral surrounding buildings and scatteredobjects (trees, cars, benches, bicycles, etc). Four

Figure 2. Figure 2 indicates the location of the pathway (grey) surrounding the test building (black). A set of target locations (A–D) are

illustrated, which were counterbalanced for orientation and so on (see text). Other grey areas represent buildings. The dashes represent a

hedge. Open areas represent spaces that contained features such as trees, benches, and so on; such objects were also scattered throughout the

virtual environment (VE). In the prerecorded tours (Experiments 2 and 3), participants explored the open areas.



Dow

nloa

ded

by [

Uni

vers

ity o

f C

alif

orni

a, S

an F

ranc

isco

] at

01:

14 3

0 Se

ptem

ber

2014

“target” objects (e.g., a bucket and a trolley) wereplaced along the explored route. The impover-ished–outdoor VE had the same central buildingbut no other buildings or objects except for thetargets; thus the surrounding area was completelybarren.

The rich–indoor environment was a corridorstructure with comparable dimension to that ofthe building surrounds in the “outside environ-ments”. The width of the corridor was 3 metres,and the environment contained several objects(doors, radiators, shelves, chairs, etc). The targetobjects (the same as those used in the outdoorenvironments) were placed around the inner wallin the same spacing and arrangement as in theoutdoor environments. The impoverished–indoorVE was a replication of the rich–indoor VE, butcontained only the target objects (the doors, radia-tors, etc. were removed).

ProcedureThe experiment consisted of three phases: fam-iliarization, exploration, and test. In the familiariz-ation phase, participants practised virtualnavigation in the practice VE, and they familiar-ized themselves with the JRD pointing task (seebelow).

The exploration phase began when participantsstarted exploring the U-shaped route. They werepresented with one of eight variations of the routein each VE. The routes differed in terms of targetobjects (four objects were chosen from a pool ofsix objects), route length (four routes had the firstleg as the shortest, and four routes had the lastleg as the shortest), and turn direction (on theoutward journey, four routes had only 90° leftturns, and four routes had only 90° right turns).

Participants moved along the route for a total ofnine times (each: start to end, returning end tostart). On the first three occasions, the exper-imenter verbally directed participants, pointingout and naming the target objects as they wereencountered. Then the participants retraced theroute a further three times (start to end, end tostart) while naming the objects themselves. Thiswas required to ensure they were attending to thetargets. On the final three traversals, neither

the experimenter nor the participants named theobjects. It is critical to emphasize that traversingthe routes from start to end as well as from endto start ensured that participants travelled alongall segments of the routes equally often in bothdirections (e.g., on the A–B segment they travellednine times from A to B and nine times from B toA). As the route was U-shaped, the orientationdefined by the (parallel) A–B and C–D route seg-ments was experienced more often than the orien-tation defined by the B–C section.

In the test phase, participants’ spatial knowledgewas assessed in a computer-based JRD task inwhich participants were presented with an outlineof a circle on the computer screen. They wereasked to imagine being located at the centre ofthe circle. A line extending from the centre of thecircle upward represented the imagined facingdirection. Task instructions printed at the top ofthe screen (“Imagine you are standing atX. Directly ahead of (behind) you is Y. Point toZ”) advised participants to indicate the directionof a target. Spatial judgements were made bymoving the centre line to point in the appropriatedirection (using the computer mouse). Theprogram recorded both decision latency and angleerror. A similar methodology has been employedin related studies (e.g., Tlauka, Donaldson, &Wilson, 2008), and in our laboratory we havefound results comparable to those obtained withtasks relying on joysticks and pointing devices.The participants were required to make 12 judge-ments, comprising 4 aligned with the first perspec-tive, 4 misaligned with the first perspective (by 90degrees), and 4 contra-aligned with the first per-spective (misalignment by 180 degrees) trials (seeWilson et al., 2007, for a test procedure that alsorelied on four trials per imagined heading). Thefirst perspective alignment was defined withrespect to the first direction of travel duringexploration. The trials comprised a mixture of sixfront-facing and six back-facing trials, with 6 jud-gements on the first and 6 on the last parts of theroute. Front-facing meant that if the route wasrecalled correctly, the test object, Z, should berecalled as forward of an imaginary line crossingthe body from left to right.


TLAUKA ET AL.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

alif

orni

a, S

an F

ranc

isco

] at

01:

14 3

0 Se

ptem

ber

2014

Results and discussion

Absolute directional errors in the JRD task (theabsolute difference between the true direction ofthe test objects and participants’ directional esti-mates) and response latencies were analysedemploying separate mixed 4× 3× 2× 2 analysesof variance (ANOVAs), with group (rich–indoor,rich–outdoor, impoverished–indoor, impover-ished–outdoor) as the between-participants factorand alignment (aligned, misaligned, and contra-aligned with the first direction of travel), judgementdirection (correct directional response: front versusback), and route section (first or last on the start toend journey) as within-group factors.

The mean absolute errors scores for the groupsare summarized in Figure 3, from which it can beseen that the FPA effect was present in allgroups. The analysis of error scores confirmed thepresence of the overall FPA effect, F(2, 184)=15.03, MSE= 2,388.35, p, .001, ηp

2= .14.Planned comparisons indicated that aligned judge-ments were more accurate than misaligned esti-mates, which, in turn, were more accurate thancontra-aligned judgements. In addition, the maineffect of judgement direction, F(1, 92)= 12.00,MSE= 1,603.82, p, .01, ηp

2= .11, the two-way

interaction between alignment and judgementdirection, F(2, 184)= 5.11, MSE= 1,625.50,p, .01, ηp

2= .05, and the three-way interactionbetween alignment, route section, and judgementdirection, F(2, 184)= 4.31, MSE= 1,247.85,p, .02, ηp

2= .04, were significant. Estimates toobjects that were front-facing with respect to thecorrect imagined test orientation were more accu-rate than back-facing estimates for all comparisonswith the exception of contra-aligned trials on thelast route section. The three-way interaction didnot reveal effects that were relevant to the hypoth-eses tested here and is not discussed further.

An additional analysis examined whether par-ticipants’ performance in the JRD task (the meanof all directional responses) was above chancelevel. Chance performance was defined as 90degrees (the range of absolute error scores was180 degrees, and chance performance was halfthis range). The analysis revealed that directionaljudgements (mean= 56 degrees) were significantlybetter than those expected by chance alone, t(95)=12.50, p, .001, providing evidence of significantlearning.

Table 1 shows a summary of the mean responselatency scores for the four groups. The main effect

Figure 3. Mean absolute error scores (in degrees) as a function of group (rich–indoor, rich–outdoor, impoverished–indoor, impoverished–

outdoor) and alignment. Error bars represent +1 estimated standard error of the mean. JRD= judgement of relative direction.



Dow

nloa

ded

by [

Uni

vers

ity o

f C

alif

orni

a, S

an F

ranc

isco

] at

01:

14 3

0 Se

ptem

ber

2014

of alignment, F(2, 184)= 11.83, MSE= 106.23,p, .001, ηp

2= .11, was found to be statisticallyreliable. Aligned trials were faster than misalignedand contra-aligned ones, which did not differ sig-nificantly. Response times were found to be mar-ginally faster on the first than on the last routesection, F(1, 92)= 3.77, MSE= 92.78, p, .06,ηp2= .03. The main effect of judgement direction,F(1, 92)= 16.12, MSE= 125.66, p, .001,ηp2= .14, and the interaction between alignmentand judgement direction, F(2, 184)= 4.05,MSE= 83.09, p, .02, ηp

2= .04, were also signifi-cant. Latencies were faster for front-facing thanback-facing judgements on aligned and misalignedtrials, with a less pronounced effect in the samedirection on contra-aligned trials.

Evidence for an FPA effect was found in all fourgroups: Performance in the JRD task was found tobe more efficient when participants adopted animagined orientation aligned with the first part ofthe route than when they adopted an orientationthat was misaligned or contra-aligned with thefirst part of the route. In addition, there was evi-dence of superior recall for judgements from ima-gined front-facing than back-facing orientations.The consistent FPA effect in all groups suggeststhat the first perspective on a novel environmentplays an important role in spatial memory. Aninteresting additional finding from Experiment 1concerned the generally superior performance on

the first route section by comparison with the last.This effect suggests that either the first part ofthe route was particularly memorable or the lastroute segment was not particularly memorable.Note that all parts of the route were exploredequally often and from both directions. A similarresult was reported in Experiment 3 of Wilsonet al. (2007).

No support was found for our hypothesis basedon the outcomes found byWilson et al. (2007). Wepredicted that the FPA effect would be mostevident in the impoverished–indoor environment,and that exploration of a rich–outdoor environmentwould lead to the greatest attenuation of the FPAeffect. To the contrary, examination of Figure 3shows that, numerically, the FPA effect was slightlygreater following exploration in the rich–outdoorVE than in the impoverished–indoor VE.

The outcome of Experiment 1 is decisive inrejecting the original hypothesis that the numberand arrangement of features and landmarks deter-mines the extent to which the FPA will be found(within the parameters investigated). However,this leaves unanswered the question of why explora-tion of the outdoor environment in Wilson et al.(2007, Experiment 3) led to an attenuated FPAeffect. The environment employed by Wilsonet al. (Experiment 3) and the rich–outdoor VE pre-sented in Experiment 1 share comparable featuresin terms of detail, complexity, and general layout.

Table 1. Response latencies in Experiments 1–3 as a function of group and alignment

Experiment Group Aligned Misaligned Contra-aligned

Experiment 1 Rich–indoor 17.10 (1.3) 19.8 (2.1) 19.8 (1.7)

Rich–outdoor 15.5 (1.3) 20.7 (2.1) 17.9 (1.7)

Impoverished–indoor 17.1 (1.3) 19.6 (2.1) 18.3 (1.7)

Impoverished–outdoor 16.4 (1.3) 20.2 (2.1) 19.7 (1.7)

Experiment 2 VE-A 13.0 (1.2) 16.6 (1.4) 15.7 (1.5)

VE-B 14.9 (1.2) 17.4 (1.4) 18.5 (1.5)

Experiment 3 Tour immediate 13.9 (0.9) 13.8 (1.0) 15.0 (1.0)

No tour 13.8 (0.9) 15.8 (1.0) 15.3 (1.0)

Tour delay 13.0 (0.9) 15.7 (1.0) 15.0 (1.0)

Note: Response latencies in seconds. Estimated standard errors are shown in parentheses. VE-A=Virtual Environment A. VE-B=Virtual Environment B.


TLAUKA ET AL.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

alif

orni

a, S

an F

ranc

isco

] at

01:

14 3

0 Se

ptem

ber

2014

The main difference between the environments wasthat Wilson et al. asked participants to explore areal environment, while the present study employedcomputer-simulated VEs. Spatial simulations havebeen used in related research (e.g., Kelly &McNamara, 2008), and we therefore consider itunlikely that this factor was the main contributorto the dissimilar outcomes.

We note that when participants in Wilson et al.(2007) were introduced to the real or recorded cor-ridor-like spaces in their Experiments 1 and 2,these routes were in novel indoor settings.However, for their Experiment 3, they selected abuilding “with which the participants were unlikelyto be familiar, since it was situated on the peripheryof the campus and three of the adjacent paths didnot lead to other campus buildings” (p. 1440).Therefore, even though participants were unlikelyto have explored the route around this building,its general situation and the surrounding campuswould have been familiar. In Experiment 2, weinvestigated whether familiarity with the area sur-rounding and including a route of similar layout tothat used in Wilson et al. (2007, Experiment 3)could have attenuated the FPA effect that is typicallyfound in more novel environments.

EXPERIMENT 2

Experiment 2 comprised two groups: Experimentalparticipants were passively shown a prerecordedtour of the virtual area that contained the to-be-explored route, prior to exploration of that route.Control participants also experienced a passivetour, but of an unrelated VE prior to explorationof the same route. We refer to the VE that con-tained the U-shaped test route as VirtualEnvironment A (VE-A). Participants in bothgroups explored this route, identifying targetobjects, and were subsequently given JRD testingrelated to this route. Prior to exploration, the firstgroup (VE-A) was given a passively experiencedprerecorded tour of the area surrounding the testroute in which VE-A was situated. The secondgroup (VE-B) experienced a similarly structuredprerecorded tour of an unrelated environment for

the same period of time, VE-B, prior to explorationof the test route in VE-A. Thus the tested routewas completely novel for only group VE-B. If fam-iliarity with the area surrounding the route attenu-ates the FPA effect, that effect should be evident ingroup VE-B but attenuated in group VE-A.

Method

ParticipantsThe participants were 48 undergraduate studentswho took part in partial fulfilment of a courserequirement. Their mean age was 21 years (range:18–39 years). They were randomly allocated totwo groups (n= 24).

ApparatusTwo VEs were employed: VE-A was the same asthat employed in the rich–outdoor condition ofExperiment 1; this VE was based on the real-world campus used in Wilson et al. (2007,Experiment 3). VE-B comprised the frames ofthe buildings from VE-A reorganized in differentlocations, but included none of the objects fromVE-A. In VE-B only a single texture was appliedto the surface of the buildings, while in VE-Aseveral textures were used. The computerizedJRD task was identical to the one employed inthe previous experiment. Several prerecordedtours were recorded in VE-A and VE-B, all wereof similar duration and complexity: These lastedbetween 165 and 246 seconds. Each tour beganwith a rapid rotation from a point within the VE,and slowed; this avoided exposure to an alternative“first perspective” on VE-A to that encounteredduring the subsequent exploration phase. Thetour comprised a variety of forward, backward,and sideways movements, which for group VE-Aincluded aspects to the to-be-explored route andthe immediately surrounding area.

ProcedureThe experiment consisted of four phases: familiar-ization, preliminary tour, exploration, and test. Thefamiliarization phase was identical to that inExperiment 1. In the tour phase, the groups (VE-A, VE-B) were passively shown a recording of a



Dow

nloa

ded

by [

Uni

vers

ity o

f C

alif

orni

a, S

an F

ranc

isco

] at

01:

14 3

0 Se

ptem

ber

2014

tour around their respective environments, VE-Aor VE-B. Then all participants actively exploredthe U-shaped route around the central buildinglocated in VE-A followed by the alignment test.The exploration and test phases were identical tothose in Experiment 1. Participants moved alongthe routes for a total of nine times (start to end,and end to start). In the JRD task, spatial knowl-edge was tested from aligned, misaligned, andcontra-aligned orientations. Counterbalancing notmentioned was the same as that in Experiment 1.


Absolute error scores and latencies in the JRD taskwere analysed employing a mixed 2× 3× 2× 2ANOVA, with group (VE-A, VE-B) as thebetween-participants factor and alignment, judge-ment direction, and route section as within-partici-pants factors. Figure 4 presents a summary of themean absolute error scores for both groups. Thestatistical analysis revealed a significant interactionbetween group and alignment, F(2, 92)= 4.28,MSE= 1,936.05, p, .02, ηp

2= .08. Recalled per-spectives that were aligned, misaligned, andcontra-aligned with the first perspective on theexplored route did not differ for group VE-A,

while for group VE-B, aligned trials were recalledsignificantly more accurately by comparison withmisaligned and contra-aligned trials, which didnot differ from each other. The main effect of jud-gement direction, F(1, 46)= 4.52, MSE=2,151.94, p, .04, ηp

2= .08, as well as the inter-action between judgement direction and alignment,F(2, 92) = 4.67, MSE = 1,187.94, p , .02,ηp2= .09, were statistically reliable. Judgementsthat were front-facing with respect to the correctimagined orientation were more accurate thanback-facing judgements on aligned and misalignedtrials. On contra-aligned trials, front and back esti-mates were comparable. The reason for the differ-ence in performance on aligned and misalignedtrials is not immediately obvious. However, a pre-ference of front over back judgements has beeninterpreted as evidence of egocentric encoding(Sholl, 1987), and based on the present results itmay be that aligned and misaligned trials weremore likely to be recalled from an egocentricperspective.

An additional analysis examined whether par-ticipants’ performance in the JRD task (the meanof all directional responses) was above chancelevel. The analysis revealed that performance(mean= 50 degrees) was significantly better than

Figure 4. Mean absolute error scores (in degrees) as a function of group (Virtual Environment A, VE-A; Virtual Environment B, VE-B) and

alignment. Error bars represent +1 estimated standard error of the mean. JRD= judgement of relative direction.


TLAUKA ET AL.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

alif

orni

a, S

an F

ranc

isco

] at

01:

14 3

0 Se

ptem

ber

2014

that expected by chance alone, t(47)= 12.27,p, .001.

In the analyses of latencies (see Table 1 for asummary), a significant main effect of alignmentwas found, F(2, 92)= 9.56, MSE= 64.27, p ,.001, ηp

2= .17. Latencies on aligned trials werefaster than those on misaligned and contra-aligned trials (individual analyses of latencies forgroups VE-A and VE-B confirmed a significantalignment effect in both groups). Front-facing jud-gements were found to be faster than back-facingjudgements, F(1, 46)= 14.67, MSE= 54.83, p ,.001, ηp

2= .24, and the three-way interactionbetween group, alignment, and route section wasfound to be significant, F(2, 92)= 5.91, MSE=34.54, p , .01, ηp

2= .11. Front-facing judgementswere faster than back-facing judgements with theexception of aligned trials in group VE-A and mis-aligned trials in group VE-B.

As in Experiment 1, front-facing judgementswere generally better than back-facing judgementsin Experiment 2. The main finding of interest con-cerns the magnitude of the FPA effect in the twogroups. With respect to response accuracy, anFPA effect was obtained for group VE-B forwhom the explored route was completely novel,but not for group VE-A for whom the route hadsome familiarity. The results are in agreementwith the hypothesis that familiarity with the areasurrounding the to-be-explored route can attenuatethe influence of the first-experienced perspective.Intriguingly, analysis of response latenciessuggested an FPA effect in both groups. APearson correlation was computed to assess therelationship between mean scores (across aligned,misaligned, and contra-aligned trials) for responselatencies and response errors. There was a positivecorrelation between the two variables (r= .301,n= 48, p, .04), suggesting no speed–accuracytrade-off (separate correlations for aligned, misa-ligned, and contra-aligned trials also revealed posi-tive correlations between error and reaction time,RT, scores).

The first perspective experienced in a novelenvironment influenced recall in bothExperiments 1 and 2, providing evidence ofstrong (possibly “default”) encoding based on the

first perspective. Experiment 2 provides supportin error scores (but not latencies) for the hypothesisthat the influence of the first perspective can beattenuated if participants have some familiaritywith the area that contains a to-be-exploredroute. This latter finding suggests that the attenu-ation of the FPA in Wilson et al. (2007,Experiment 3) occurred because their participantshad some familiarity with the surroundingcampus. However, it is noteworthy that Wilsonet al. found attenuation of the FPA in both errorsand latencies in their Experiment 3. Possibly, therelatively brief preexposure to the area surroundingthe route that was experienced by group VE-Acould have been insufficient to achieve the morecomplete attenuation of the FPA effect thatextended experience of the campus might haveengendered in Wilson et al.’s participants.

EXPERIMENT 3

The robustness of the FPA effect in Experiment 1,together with the examples from the literaturedocumented in the introduction, suggests that theinfluence of the first perspective on referenceframe formation is strong and pervasive. Theattenuation of the FPA effect found inExperiment 2 could have been the consequence ofmultiple views of the environment that containedthe to-be-explored route. According to thisinterpretation, preliminary experience establishedan allocentrically based spatial representation intowhich the explored route was subsequently incor-porated. Essentially, the prerecorded tour of thearea led to encoding of the whole space in asingle allocentric representation with no consist-ently preferred orientation (i.e., across significantnumbers of participants).

Alternatively, the memory of the tour of the sur-rounding area could have been established quiteindependently from that of the explored route,with the explored route still encoded primarily withreference to the first perspective. In this case, theFPA effect was attenuated in Experiment 2 by thetour experience because at the time of testing, mem-ories of the surrounding area “masked” the influence



Dow

nloa

ded

by [

Uni

vers

ity o

f C

alif

orni

a, S

an F

ranc

isco

] at

01:

14 3

0 Se

ptem

ber

2014

of the preferred first perspective on the exploredroute (rather than attenuating its formation).

In Experiment 3, a delay between learning andtesting was introduced to help assess these alterna-tives. Tlauka et al. (2008; see also Elmes, 1988)found that spatial memories of virtual spaces aredetrimentally affected by a one-week period. If pre-ferential encoding of the first perspective is alwaysestablished during exploration of a novel route,but can be masked in the short term by a memoryof the preceding tour, the influence of the tourmight weaken during a one-week delay as lessmemorable details of the surroundings fade frommemory. The “reemergence” of an FPA effectafter a delay, particularly if the effect is of asimilar magnitude to that found with no tour ofthe surroundings, would be consistent with prefer-ential encoding of the first perspective at the timethe novel route was explored.

Alternatively, if a single allocentric represen-tation was formed that comprised the exploredroute integrated into the surroundings, reencodingof that route following a delay would not be antici-pated. Attenuation of the FPA effect with both animmediate test and a delayed test would stronglysuggest that a single memory (in the form of anamalgamation of preliminary tour and exploredroute) was established at encoding.

Three groups of participants were tested. Twogroups, “tour immediate” and “tour delay” wereshown a tour of the area surrounding the to-be-explored route in VE-A. Following the tour, bothgroups explored the U-shaped routes around thecentral building. Group tour immediate then tookpart in JRD assessment, while for group tourdelay, the JRD assessment was administered sevendays later. A third group (no tour) explored the U-shaped route and was given the alignment taskimmediately after exploration, but this group wasnot presented with a prerecorded tour and was there-fore predicted to show a “normal” FPA effect.

Method

ParticipantsThe participants were 144 undergraduate studentswho took part in partial fulfilment of a course

requirement. Their mean age was 26 years (range:18–57 years). They were randomly allocated tothree groups (n= 48).

ApparatusThe virtual environment (VE-A) and the compu-terized JRD task were the same as those employedin Experiment 2.

ProcedureThe experiment consisted of four phases: familiar-ization, tour, exploration, and test. For all groups,the familiarization phase was identical to that inExperiments 1 and 2. In the tour phase, groupstour immediate and tour delay were shown arecorded tour of VE-A. They then explored theU-shaped route around the central buildingwithin VE-A. Participants repeatedly negotiatedthe designated routes, and in the subsequent JRDtask they were tested from aligned, misaligned,and contra-aligned orientations with respect tothe first perspective. Group tour immediate wastested immediately after exploration of the routes.Group tour delay was tested seven days later.Group no tour was treated identically to grouptour immediate with the exception of the tourphase. Participants in this group did not view atour of VE-A prior to exploration of the testroute. Counterbalancing was the same as that inExperiment 2.


Absolute error scores and latencies in the JRD taskwere analysed employing a mixed 3× 3× 2× 2ANOVA, with group (tour immediate, no tour,tour delay) as the between-participants factor andalignment, judgement direction, and route sectionas within-participants factors.

Error scores are illustrated in Figure 5. A sig-nificant main effect of alignment was obtained,F(2, 282)= 14.34, MSE= 2,504.80, p , .001,ηp2= .09, which was qualified by a significantinteraction between group and alignment, F(4,282)= 2.51, MSE= 2,504.80, p , .05, ηp

2= .03.Interaction contrasts indicated that for groupstour delay and no tour, responses were more


TLAUKA ET AL.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

alif

orni

a, S

an F

ranc

isco

] at

01:

14 3

0 Se

ptem

ber

2014

accurate on aligned trials than on contra-alignedtrials. No other differences were significant. Forgroup tour immediate, response accuracy did notdiffer on aligned, misaligned, and contra-alignedtrials. The analysis further revealed more accuratefront than back judgements, F(1, 141)= 7.12,MSE= 1,968.17, p , .01, ηp

2= .04. There wasalso a significant three-way interaction betweenalignment, judgement direction, and routesection, F(2, 282)= 4.42, MSE= 1,552.78, p ,.02, ηp

2= .03, which did not reflect on the hypoth-esis under test.

An additional analysis examined whether par-ticipants’ performance in the JRD task (the meanof all directional responses) was above chancelevel. The analysis revealed that performance(mean= 55 degrees) was significantly better thanthat expected by chance alone, t(143)= 16.83,p, .001.

With respect to the latency scores (see Table 1),the main effect of alignment was found to be sig-nificant, F(2, 282)= 9.63, MSE= 47.71, p ,.001, ηp

2= .06, and the interaction between groupand alignment was marginally significant, F(4,282)= 2.28, MSE= 47.71, p = .06, ηp

2= .03. Asin Experiment 2, some evidence of the FPA

effect was suggested in the latency scores of allgroups, but with a closer resemblance between thepatterns for groups no tour and tour delay thanfor tour immediate. The analysis further indicatedthat latencies on the first route section were fasterthan those on the last section, F(1, 141)= 5.76,MSE= 40.79, p , .02, ηp

2= .03, and responsetimes were faster for front-facing than for back-facing judgements, F(1, 141)= 16.41, MSE=84.31, p , .001, ηp

2= .10. The four-way inter-action between group, alignment, judgement direc-tion, and route section also reached statisticalsignificance, F(4, 282)= 3.67, MSE= 49.25, p ,.01, ηp

2= .04, but is not related to any of thehypotheses tested in Experiment 3.

Experiment 3 replicates the effect found inExperiment 2. A passive tour of the area sur-rounding the to-be-explored novel route wasfound to attenuate the typical influence of thefirst perspective on recall of the explored route.This pattern relied on the JRD testing beingadministered immediately after exploration.Delaying the recall test for seven days led to apattern of judgement efficiency remarkablysimilar to that of people who had no experienceof the passive tour.

Figure 5. Mean absolute error scores (in degrees) as a function of group (tour immediate, no tour, tour delay) and alignment. Error bars

represent +1 estimated standard error of the mean. JRD= judgement of relative direction.



Dow

nloa

ded

by [

Uni

vers

ity o

f C

alif

orni

a, S

an F

ranc

isco

] at

01:

14 3

0 Se

ptem

ber

2014

The results of Experiment 3 suggests that anexplored route is always encoded giving priority tothe first perspective or direction of travel. Prior allo-centric experiencing of the surrounding space canmask the influence of the first perspective in theimmediate term. The findings are consistent withthe assumption that the source of the FPA effectis at encoding, while expression of the FPA canbe influenced at recall.

GENERAL DISCUSSION

This investigation examined the conditions thatpromote a frame of reference that is tied to thefirst perspective or direction of travel within anovel environment. In three experiments con-ducted in VEs, evidence was found of most efficientrecall of spatial relationships that were in alignmentwith the first perspective experienced duringexploration. In Experiment 1, this preferential effi-ciency in performance was independent of whetherlearning occurred in a restricted (barren corridors)or rich (more typical of everyday urban life)environment. These FPA effects are in agreementwith results reported in related studies (Palijet al., 1984; Richardson et al., 1999; Rossanoet al., 1999; Wilson et al., 1999; Wilson &Wildbur, 2004; Wilson et al., 2007). The resultsraise doubts about the claim (Mou et al., 2007)that the initial viewing perspective is not criticalin the selection of a preferred reference direction(note, however, that Mou et al.’s findings were inthe context of highly structured layouts).

Experiment 2 identified a variable that attenu-ated the FPA effect as assessed by directional jud-gement accuracy. A few minutes exposure to apassive tour of the area containing the to-be-explored route attenuated the influence of the firstperspective on that route in determining the accu-racy with which subsequent directional judgementscould be made. Two theoretical frameworks wereconsistent with the findings. First, attenuation ofthe FPA effect suggests encoding of the routeand surroundings in a single allocentric formatwith no consistent preferred orientation.

Second, the results from Experiment 2 could beexplained in terms of multiple mental represen-tations, with the tour of the explored route beingrepresented independently from the surroundingarea. According to this view, the explored routecould have been represented relative to the first per-spective, while the surrounding area was rep-resented in a less consistently preferredorientation. This interpretation assumes thatexperience of the recorded tour attenuated theFPA effect in Experiment 2 because at recall thememories of aspects to the surrounding area“masked” the encoding of the preferred firstperspective.

Experiment 3 contrasted these rival accountsand found that delaying JRD testing led to the ree-mergence of the influence of the first perspective.This finding is consistent with the influence ofthe tour experience weakening during the one-week delay as less well-processed details of the sur-roundings faded from memory. The return of anFPA effect after a delay is in agreement with thehypothesis that the first perspective was preferen-tially encoded at the time the novel route wasexplored. Had the first perspective not beenspecially encoded at the time of exploration, it isdifficult to envisage how it could have reemergedafter a delay.

With sufficient ingenuity it is possible todevelop an alternative account that invokes recallas the source of the FPA effect, but we considerthis much less plausible. If we suppose that thesource of the FPA is at recall, this suggests thatwhen asked to make directional judgements,people call up a spatial memory that has essentiallyno preferred orientation, and at this stage theyadopt a perspective that is aligned with the firstexperienced perspective during exploration. Thefirst source of implausibility is that there is littlereason to adopt this perspective unless it was par-ticularly memorable during exploration (i.e., atencoding). The second is that while the first“view” (e.g., A to B in Figure 1) might be particu-larly memorable due to a “primacy effect”, the FPAalso influences perspectives experienced later in thecourse of exploration that are in the same alignment(D to C in Figure 1).


TLAUKA ET AL.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

alif

orni

a, S

an F

ranc

isco

] at

01:

14 3

0 Se

ptem

ber

2014

It is noteworthy that in agreement with pre-vious research (Sholl, 1987) front-facing judge-ments were superior to back-facing judgements.This suggests an egocentric perspective at recall,“as if tested within the environment”. If peoplewere to imagine the route from a map-like per-spective, the direction of objects to the front andrear of the observer’s imagined location shouldbe recalled equally efficiently. Several models ofspatial memory are based on a distinctionbetween egocentric and allocentric representations(Mou et al., 2004; Sholl & Nolin, 1997; Waller &Hodgson, 2006). Given the present findings andthose obtained in related studies, our interpret-ation of the literature is in agreement with thosetheories that postulate that egocentric referencesystems are an essential component of spatial rep-resentations. Note, however, that the FPA effectreflects an interaction between egocentric andallocentric influences on spatial representations.With reference to Figure 1, the initial egocentricperspective on the A→ B segment promotesmore efficient recall of that view, but it alsopromotes more efficient recall of D→C.During each exploration of the route, C→Dwas encountered before D→C (greaterefficiency for D→C judgements argues againstparticipants learning a spatial list of the form A,B, C, D, C, B, A, which predicts that C→Dshould be better learned than D→C because itis earlier in the list). Therefore, egocentricexposure at A→ B must establish a “conceptualnorth” that orients the entire environment, withmore efficient C→D judgements because thatperspective is in alignment with overall allocentricencoding.

In summary, the FPA effect is a strong and con-sistent feature of spatial memory for exploredroutes. It appears to be established at encoding,and although familiarity with the area that containsthe route can weaken the effect in the short term,with our parameters it reemerges with the passageof time.

Original manuscript received 8 September 2010

Accepted revision received 20 April 2011

First published online 15 August 2011

REFERENCES

Burgess, N. (2006). Spatial memory: How egocentric andallocentric combine. Trends in Cognitive Sciences, 10,551–557.

Couclelis, H., Golledge, R. G., Gale, N., & Tobler, W.(1987). Exploring the anchor-point hypothesis ofspatial cognition. Journal of Environmental

Psychology, 7, 99–122.Easton, R. D., & Sholl, M. J. (1995). Object-array struc-

ture, frames of reference, and retrieval of spatialknowledge. Journal of Experimental Psychology:

Learning, Memory and Cognition, 21, 483–500.Elmes, D. G. (1988). Interference in spatial memory.

Journal of Experimental Psychology: Learning,

Memory & Cognition, 14, 668–675.Franklin, N., & Tversky, B. (1990). Searching imagined

environments. Journal of Experimental Psychology:

General, 119, 63–76.Hegarty, M., Montello, D. R., Richardson, A. E.,

Ishikawa, T., & Lovelace, K. (2006). Spatial abilitiesat different scales: Individual differences in aptitude-test performance and spatial-layout learning.Intelligence, 34, 151–176.

Kelly, J. W., & McNamara, T. P. (2008). Spatial mem-ories of virtual environments: How egocentricexperience, intrinsic structure, and extrinsicstructure interact. Psychonomic Bulletin & Review,15, 322–327.

Mou, W., McNamara, T. P., Valiquette, C. M., &Rump, B. (2004). Allocentric and egocentric updat-ing of spatial memories. Journal of Experimental

Psychology: Learning, Memory and Cognition, 30,

142–157.Mou, W., Zhao, M., & McNamara, T. P. (2007).

Layout geometry in the selection of intrinsic framesof reference from multiple viewpoints. Journal of

Experimental Psychology: Learning, Memory and

Cognition, 33, 145–154.Palij, M., Levine, M., & Kahan, T. (1984). The orien-

tation of cognitive maps. Bulletin of the Psychonomic

Society, 22, 105–108.Richardson, A. E., Montello, D. R., & Hegarty, M.

(1999). Spatial knowledge acquisition from maps,and from navigation in real and virtual environments.Memory & Cognition, 27, 741–750.

Rossano, M. J., West, S. O., Robertson, T. J.,Wayne, M. C., & Chase, R. B. (1999). The acqui-sition of route and survey knowledge from compu-ter models. Journal of Environmental Psychology, 19,101–115.



Dow

nloa

ded

by [

Uni

vers

ity o

f C

alif

orni

a, S

an F

ranc

isco

] at

01:

14 3

0 Se

ptem

ber

2014

Shelton, A. L., & McNamara, T. P. (2001). Systems ofspatial reference in human memory. Cognitive

Psychology, 43, 274–310.Sholl, M. J. (1987). Cognitive maps as orienting sche-

mata. Journal of Experimental Psychology: Learning,

Memory and Cognition, 13, 615–628.Sholl, M. J., & Nolin, T. L. (1997). Orientation speci-

ficity in representations of space. Journal of

Experimental Psychology: Learning, Memory and

Cognition, 23, 1494–1507.Tlauka, M., Donaldson, P., & Wilson, D. (2008).

Forgetting in spatial memories acquired in a virtualenvironment. Applied Cognitive Psychology, 22, 69–84.

Waller, D. (2000). Individual differences in spatial learn-ing from computer-simulated environments. Journalof Experimental Psychology: Applied, 6, 307–321.

Waller, D., & Hodgson, E. (2006). Transient andenduring spatial representations under disorientationand self-rotation. Journal of Experimental Psychology:Learning, Memory and Cognition, 32, 867–882.

Wilson, P. N., Foreman, N., & Tlauka, M. (1997).Transfer of spatial information from a virtual to areal environment. Human Factors, 39, 526–531.

Wilson, P. N., Tlauka, M., & Wildbur, D. J. (1999).Orientation specificity occurs in both small- andlarge-scale imagined routes presented as verbaldescriptions. Journal of Experimental Psychology:

Learning, Memory and Cognition, 25, 664–679.Wilson, P. N., & Wildbur, D. J. (2004). First-perspec-

tive alignment effects in a computer-simulated streetenvironment. British Journal of Psychology, 95,

197–217.Wilson, P. N., Wilson, D. A., Griffiths, L., & Fox, S.

(2007). First-perspective spatial alignment effectsfrom real-world exploration. Memory & Cognition,35, 1432–1444.

Witmer, B. G., Bailey, J. H., & Knerr, B. W. (1996).Virtual spaces and real world places: Transfer ofroute knowledge. International Journal of Human-

Computer Studies, 45, 413–428.


TLAUKA ET AL.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

alif

orni

a, S

an F

ranc

isco

] at

01:

14 3

0 Se

ptem

ber

2014

the first-perspective alignment effect: the role of environmental complexity and familiarity with...

Documents